Startup Touts "First Optical Router"
Another startup has emerged with even more ambitious plans. It says that it can match Calient's claims on density -- cramming a 1,000 by 1,000 port switch into a box the size of a kitchen drawer -- and go one better. It claims its box will be a genuine optical router, one that shunts traffic onto different connections on a packet-by-packet basis.
So, who is this outfit?
It's Synchordia Networks, Inc., of Atlanta, Georgia, for the time being. The company is planning to change its name, to avoid confusion with Syncordia, a subsidiary of British Telecommunications PLC also based in Atlanta.
Right now, Synchordia is at an early stage of development. It was launched in August 1999, has 15 staff and is currently finalizing Series A financing with Metropolis Venture Partners (no web site). Seed finance came from Zilog, Inc http://www.zilog.com, Delta Asset Management Group (no web site) and angel investors.
Synchordia has been in stealth mode until now, but after seeing Calient's announcement on Monday, it changed its tune.
Why? Because Synchordia reckons it can go one better than Calient, by developing an optical router. Specifically, it's combining tunable lasers with a switching core made from MEMS (micro-electrical mechanical systems), arrays of tiny tilting mirrors. The tunable lasers come from Altitun AB http://www.altitun.com, a startup being acquired by ADC Telecommunications, Inc. http://www.adc.com (see ADC Scores a Coup on Tunable Lasers). Altitun's lasers can switch from one wavelength to another in a matter of nanoseconds, a short enough period of time to route individual packets onto different connections.
Whether or not this constitutes a true optical router (or photonic router) is a moot point. It's a lot closer to being the real thing than the LambdaRouter from Lucent Technologies http://www.lucent.com, (see Optical Illusions). But Synchordia's box is really only half-way towards being a genuine optical equivalent of a conventional IP router. It can't route packets that come in as light, but it can take in traffic in electrical form, convert it into light pulses and then route it over an optical backbone on a packet by packet basis.
Incoming traffic has to be in electrical form because the output from Atitun's lasers is modulated (interrupted to generate a stream of pulses) using electricity. Similarly, Synchordia's router needs to read the addresses in the incoming packets, and that also has to be done electrically at present.
In the long run, most all-optical switches will end up having to combine tunable lasers and optical switching fabric. That's because the switching fabric itself merely provides a link between one fiber and another (hence the jibes about all-optical switches being automated patch panels). Adding tunable lasers will enable traffic to be switched from one wavelength to another, massively increasing effective capacity.
The best way of understanding this is to consider a 256 x 256 port switch being used with 40 channel DWDM (dense wave division multiplexing). In this setup, each port is connected to a fiber carrying 40 wavelengths. Without tunable lasers, incoming traffic on a particular wavelength can only be switched to the same wavelength on another port - 255 options. With tunable lasers, the traffic can be routed to any wavelength on any port - 10,200 options.
A big question remains over whether Synchordia will be able to take this concept a stage further and route traffic over such a setup on a packet by packet basis.
Right now, Synchordia can't match Calient's line up of experts, which gives its project some credibility. Gerald Ramdeen, Synchordia's CEO, is the only member of staff with a high profile track record of developing networking technology - and even his experience is a little whacky. Ramdeen developed ultra-fast packet networks for the B2 stealth bomber.
Ramdeen says he's close to hiring a whole development team of "heavy hitters" from a major vendor. He's planning on having a 256 by 256 port prototype router in the fourth quarter of this year, so that he can start trials in early 2001.
-- by Peter Heywood, international editor, Light Reading http://www.lightreading.com